2006 Conference General Sessions


Marguerite Bergel
Fidelity Investments
82 Devonshire Street, V4A
Boston MA 02109
Day Phone: 617-392-2069
Fax: 617-476-0291
Email: marguerite.bergel@fmr.com

Presenter #2
Ann Chadwick-Dias
Fidelity Investments
82 Devonshire Street, V4A
Boston MA 02109
Day Phone: 617-392-1904
Email: annmarie.chadwick-dias@fmr.com

This presentation describes a study involving 21 JAWS Web users. We will explain and illustrate (with captioned video clips from user sessions) the effectiveness of various user navigation strategies.

We conducted an empirical research study involving 21 JAWS users. As we observed participants completing tasks, we noted trends in their different approaches to Web navigation including that some strategies were significantly more effective than others. To further explore this, we performed post-hoc analysis to define the different strategies and correlate them to performance. From here, we grouped similar strategies into 4 basic user interaction styles to more effectively describe and encapsulate the key differences among users’ strategies we observed. We hope this model may also help assistive technology trainers, Web designers, developers, and vendors to more effectively design their training courses, Web sites, and software to better meet the needs of blind Web users.

Participants in our study saw 2 sets of tasks – 1 for each prototype. Each task set asked users to complete an equal mix of objectives such as finding contact information or updating personal profile information.

Participants completed up to 5 tasks on each version of the prototype. We collected task time, task success, and the navigation path.

Understanding User Navigation Strategies
Understanding how blind JAWS users can, in theory, navigate a site is different than understanding how, in practice, they actually do. Even knowing the full range of JAWS commands available to users reveals little about how they will string the commands together into a larger navigation strategy when completing tasks online. For that matter, different strategies may be more effective at different stages in the process of completing a task. In this presentation, we divide Web navigation into two categories: inter-page and intra-page navigation.  

Inter-page Navigation
Inter-page navigation requires a user to find a path through the site to a content page where the desired information (e.g. contact number or a specific form) resides. This requires adopting a cognitive model of the most likely path through the site hierarchy toward that destination page. Problems occur when the user’s mental model does not match the information hierarchy of the site.

Intra-page Navigation
Once navigating within a particular page, a user must either pinpoint the exact location of their desired content or learn enough to determine that the content is not there. Such intra-page navigation may be especially challenging for blind users since it requires them to guess the format in which their desired content is represented. For example, many users in our study employed different strategies to find telephone numbers than when attempting to update their personal profile information. In updating personal information, users tended to look for form fields immediately (by pressing “F”). To find a phone number, users did not bother with links list since they knew a phone number was likely to be a static piece of alpha-numeric text.

Navigation Possibilities in JAWS and IE
For those unfamiliar, JAWS users can employ various Web navigation functions at any time, such as:
* reading all the information on a page (“read all”)
* displaying all page links in a list and searching through it via first-letter navigation or arrowing up and down (“links list”)
* jumping across like elements on a page without listening to the rest of the content (different commands exist to jump between headings, paragraphs, frames etc.)
* jumping to specific elements via hotkeys where available (shortcut or access keys)
* searching for specific words or phrases on the page (Ctrl Find)
* verifying what page they are on (Ctrl Home)

While a user theoretically could employ any of the above tactics to both inter- and intra- page navigation, preliminary analysis suggests certain strategies are most effective for each. We will describe the more effective navigation strategies for both inter- and intra- page navigation at length and support them with captioned video clips of users from our study.

Preliminary Results
Preliminary analysis indicates that those most effective in completing tasks (time per task and task completion) using JAWS:
* used headings to learn site and page-level structure
* accessed additional structural elements like frames and tables to learn site and page-level structure
* avoided selecting the first link that seemed promising (especially important given the large number of links per page on prototypes in this study)
* spent less time “getting a sense” for the home page prior to performing tasks
* employed a wider range of inter-page and intra-page navigation strategies when seeking discrete information
* were more effective in leveraging prior knowledge of other sites’ structure and content when “envisioning” how our prototypes might work

Web Interaction Profiles
We observed 4 primary interaction strategies employed by users as they completed tasks. A full description of the characteristics of each, along with associated performance metrics, will be discussed. Below is an outline of one such profile based on preliminary analysis.

Example Profile
Users in this particular profile tended to:
* approach the site with the belief that the information they seek is there.
* theorize, experiment, and use deductive reasoning at each step within the task.
* be goal-oriented and ignore content un-related to their immediate goal.
* spend comparatively little time listening to the home page in favor of scanning it to learn its overall organization and content.
* employ multiple strategies to get a basic sense of how the site and pages were structured.
* focus on specific page content as a last step.
* Complete tasks in relatively short segments of time with high task success.  

While a good deal of research investigating Web navigation strategies of sighted users exists, comparatively little has been done that helps to describe Web interaction models for blind users. Such interaction models would paint richer, more nuanced portraits of different Web navigation strategies than simply thinking of them in terms of keyboard commands and sequences. Such interaction models could prove useful in teaching users screen reader software, establishing personas for user-centered design techniques, and designing assistive technology products.

We learned a lot about different JAWS user Web navigation strategies, the disparity between them and their effectiveness over time. Many participants we observed in the lab tended to adhere to a relatively narrow set of JAWS commands throughout. Thus the question remains, if JAWS users supposedly learn the wide variety Web navigation methods in training, why did they not demonstrate a broader range of strategies in our study? Furthermore, the 21 users in our study collectively possessed a great deal of valuable information. So why did they not seem to have shared their expertise with each other? Most of our users even knew each other socially. We even observed husbands and wives who failed to share significant strategies with each other.

Our previous research with older, sighted users demonstrated a strong correlation between Web expertise and time spent in collaborative learning environments (Chadwick-Dias, Tedesco, & Tullis, 2004). Community groups for the blind and visually impaired then might consider how to facilitate more collaborative learning opportunities related to the Web and assistive technologies. In our community, we have a group called Visually Impaired and Blind Computer Users Group (VIBUG), which among other things, dedicates time in its monthly meetings for users to ask technology-related questions and share expertise. Perhaps this collaborative model could be encouraged at a national level.

Because the Web largely lacks development or design standards, understanding how best to optimize a site for blind users remains a significant challenge. If each Web site is coded uniquely, users cannot stick to a consistent, effective approach as they attempt to accomplish tasks. Consider HTML headings as an example of how the lack of standards propagates usability problems for blind Web users. While scanning the headings on a page is an effective strategy for a blind user, many pages do not have them. Consequently, users cannot rely on scanning headings as a primary navigation strategy. Until more accessible Web development standards are defined and mandated, users will be forced to approach each page with a set of potential strategies that they can utilize. This is obviously not the ideal. Web site developers, AT developers, and users will need to work together to better influence the direction of Web access for the blind.

Session Style and Outline:
The presentation will be interactive and participation from the audience will be encouraged. Captioned video clips from user sessions will support presentation of findings.

Our presentations will be divided into the following 3 sections:
* Description of study method and background
* Discussion of Web navigation strategies and how each correlated to performance
* Description of the 4 primary Web interaction profiles we observed, including related performance metrics

Chadwick-Dias, A., Tedesco, D., Tullis, T. (2004). Older adults and web usability: is web experience the same as web expertise? CHI Extended Abstracts 2004: 1391-1394.  Available at: http://doi.acm.org/10.1145/985921.986072

Go to previous article
Go to next article
Return to 2006 Table of Contents

Reprinted with author(s) permission. Author(s) retain copyright